Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332584899

Seismic Fragilities of High-Voltage Substation Disconnect Switches

Article  in  Earthquake Spectra · April 2019


DOI: 10.1193/030118EQS049M

CITATIONS READS
2 230

3 authors, including:

Wen Bai Mohamed A. Moustafa


China Earthquake Administration University of Nevada, Reno
11 PUBLICATIONS   11 CITATIONS    86 PUBLICATIONS   230 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Seismic Performance of Ultra High Performance Concrete (UHPC) Bridge Columns View project

Seismic Behavior of SCBFs under Long Duration Ground Motion View project

All content following this page was uploaded by Wen Bai on 27 August 2019.

The user has requested enhancement of the downloaded file.


The Professional Journal of the Earthquake Engineering Research Institute

PREPRINT
This preprint is a PDF of a manuscript that has been accepted for
publication in Earthquake Spectra. It is the final version that was
uploaded and approved by the author(s). While the paper has been
through the usual rigorous peer review process for the Journal, it has
not been copyedited, nor have the figures and tables been modified for
final publication. Please also note that the paper may refer to online
Appendices that are not yet available.

We have posted this preliminary version of the manuscript online in


the interest of making the scientific findings available for distribution
and citation as quickly as possible following acceptance. However,
readers should be aware that the final, published version will look
different from this version and may also have some differences in
content.

The DOI for this manuscript and the correct format for citing the paper are
given at the top of the online (html) abstract.

Once the final, published version of this paper is posted online, it will
replace the preliminary version at the specified DOI.
1 Seismic Fragilities of High-Voltage Substation
2 Disconnect Switches
3 Bai Wen,a,b) M. EERI, Mohamed A. Moustafa,b) M. EERI, and Dai Junwua)

4 Electrical substations and vital components, e.g. disconnect switches, experienced


5 severe damage that caused blackouts during past earthquakes. To improve the seismic
6 resiliency of power grids and use probabilistic decision making frameworks,
7 comprehensive fragility data for the different substations equipment is needed. The
8 objective of this study is to investigate the seismic performance and develop
9 component and system fragility curves for a critical substation component: disconnect
10 switches. The seismic vulnerability of two common switch types in the Western United
11 States was evaluated and two operational configurations were considered. Detailed
12 nonlinear finite element models for the full switches were calibrated and validated
13 using previous experimental data and used to conduct incremental dynamic analysis
14 and carry out the fragility assessment. Total of 160 triaxial ground motions representing
15 four sets of different ground motion characteristics was used. The disconnect switches
16 fragility curves were developed and presented to be readily used for regional risk
17 assessments and reliability analysis of power grids to inform substations maintenance,
18 inspection, and retrofit activities.

19 INTRODUCTION

20 Power is indispensable for supporting reconnaissance efforts after natural hazards such as
21 earthquakes. Blackouts and power outage have adverse effects on the community at large due to
22 the interdependence between the power grid and other critical infrastructure such as water, gas,
23 transportation, and communications, etc. (Araneda et al. 2010). Thus, avoiding blackouts or
24 providing a robust way for immediate power restoration after earthquakes and other extreme events
25 is necessary for resilient communities. Power outage that is associated with local distribution
26 networks that is affecting small regions or neighborhoods is usually easier and quicker to address

a)
Key Laboratory of Earthquake Engineering and Engineering Vibration, Institute of Engineering Mechanics, China
Earthquake Administration, Harbin, Heilongjiang, 150080, China
b)
Department of Civil and Environmental Engineering, University of Nevada, Reno, NV, 89557, USA
27 than regional outages that are originated from electrical substations. Failure or damage of one or
28 more components of high-voltage substations can also lead to cascading failures. Evidence from
29 recent earthquakes shows that cascading failures have been tied to earthquakes as well (Schiff
30 2000; Khalvati 2009; Knight 2009; Massie 2011). For instance, two weeks after the February 27,
31 2010 Chile earthquake, i.e. on March 14th, a widespread cascading blackout took place because of
32 the disconnection of a 750 kV transformer bank that was attributed to a protection control cable
33 on the transformer that got loosened during the earthquake (Dueñas and Kwasinski 2012).
34 For addressing regional and cascading failures after earthquakes, electrical engineers have
35 been collaborating with earthquake engineers to develop probabilistic and reliability frameworks
36 for assessing the seismic vulnerability of the power grids (e.g. Vanzi 1996; Shinozuka et al. 2003;
37 Hines et al. 2009). Establishing comprehensive and robust fragility functions for the different
38 substation equipment is crucial for such regional assessment frameworks that are needed for
39 regular maintenance planning or prioritizing post-earthquake reconnaissance efforts. Limited
40 fragility functions for broad classes of electrical equipment are available (e.g. Paolacci et al. 2014;
41 European SYNER-G Reference Report: Kaynia et al. 2013; Reinhorn et al. 2011; Paolacci et al.
42 2009; Takhirov et al. 2009; Paolacci et al. 2005; Gilani et al. 1999). These fragility functions are
43 generic or empirical and usually do not account for the different operation configurations of the
44 different substation equipment. In this study, we focus on one critical type of the high-voltage
45 electrical substation equipment, namely vertical-break disconnect switches, and our goal is to
46 develop comprehensive fragility curves and demonstrate that such fragilities are sensitive to the
47 operation configuration. This study will also set the stage for future studies to develop
48 comprehensive fragility and vulnerability functions for different structural (e.g. circuit breakers)
49 and non-structural (e.g. protection and control devices) components of electrical substations.
50 Two types of disconnect switches (DS) with porcelain post insulators commonly used in
51 Western United States, 230-kV and 550-kV, that uses two different types of support structures are
52 considered in this study. The specific objectives of this study are: (1) develop detailed finite
53 element (FE) models for the two types of DS in different operation configuration, and
54 validate/calibrate the models against experimental results from material mechanical tests, static
55 pull tests, and shake table tests conducted by one of the authors (Moustafa and Mosalam 2016a,
56 2016b; Mosalam et al. 2012); (2) conduct nonlinear incremental dynamic analysis using the
57 validated FE models to develop new seismic fragility curves; (3) identify and develop component

2
58 versus overall system fragility curves to better understand modes of failure of disconnect switches
59 and use both peak ground acceleration (PGA) and peak response acceleration (PRA, which
60 represents the maximum acceleration response at concerned insulator top) as two different
61 intensity measures (IMs); and (4) study the effect of different site conditions and the characteristics
62 of ground motions on seismic fragilities of disconnect switches by considering 4 different sets of
63 ground motions, each is comprised of 40 triaxial ground motions. A brief background section is
64 first presented to review the relevant literature that focused on fragility analysis of electrical
65 equipment as it relates to the knowledge gaps addressed in this study. The background section is
66 followed by several sections to present the FE model development and validation, PGA and PRA-
67 based fragilities using all 160 sets of ground motions, system and component fragilities evaluation
68 and comparison, discussion of the effect of ground motion characteristics, and overall conclusions
69 and recommendations for future work.

70 BACKGROUND

71 The seismic performance of electrical equipment that uses porcelain post insulators or bushings
72 can be unsatisfactory and affect the overall substation response (Qiang Xie et al. 2011; Kaynia et
73 al. 2013; Moustafa and Mosalam 2016). During the 2008 Wenchuan earthquake in China, at least
74 90 substations with voltages of 110-kV and above were damaged (e.g. all equipment in the
75 Ertaishan switchyard were completely damaged) and 181 power transmission lines were
76 interrupted with the consequent power loss estimated as high as 6,627 MW (Qiang Xie et al. 2011).
77 During the 2010 Chilean earthquake, the Chilean Central Interconnected System, which supplies
78 electricity to more than 93% of the Chilean population, was damaged (Araneda et al. 2010).
79 Disconnect switches represent one type of substations equipment that have experienced frequent
80 seismic damage during past events such as the aforementioned cases. DS commonly use porcelain
81 post insulators and can have different types based on voltage capacity of the substation (e.g. 230
82 or 550 KV) and different operation configurations (e.g. opened gate to disconnect a transmission
83 line during maintenance). Several studies focused on experimental seismic qualification of DS
84 (e.g. Thornberry and Hardy 1997; Gilani et al. 2000; Takhirov 2008) but only few studies focused
85 on developing detailed fragility curves for DS.
86 Seismic Fragility analysis has been widely used in civil engineering areas to better understand
87 and quantify the probability of structure failure with different applications such as buildings (e.g.

3
88 Rota et al. 2010), bridges (e.g. Nielson 2007), and power infrastructure (e.g. Shinozuka et al. 2007;
89 Panteli and Mancarella 2015; NIBS 2003). Fragility studies vary not only in the applications, but
90 the definition of damage states, intensity measures, and methodology. According to practical
91 applications, power systems are considered at four analysis levels. These are substation
92 components (e.g., Anagnos 1999, Zareei et al. 2016), substations (e.g., Giovinazzi and King 2009,
93 Vanzi et al. 2005), distribution systems (e.g. NIBS 2003), and full networks (e.g., Chang et al.
94 2011, Dueñas et al. 2007a). Typical porcelain components used in electrical equipment such as
95 circuit breakers, arresters, current transformers, voltage transformers, and disconnect switches, all
96 belong under substation components.
97 Two different methods have been traditionally used in defining damage states and the data
98 points needed to derive probability functions, and in turn, develop the fragility curves: (1)
99 empirical methods (e.g. Straub and Der Kiureghian 2008, Dueñas et al. 2007b); and (2) numerical
100 methods (e.g. Rasulo et al. 2004). Empirical data can be acquired from earthquake damage
101 investigation and empirical judgment or from experimental tests. The shortcomings of empirical
102 methods is that is highly dependent on personal judgment and expertise, so it lacks consistency,
103 and in many cases, the events leading to the damage data collection such as earthquake
104 reconnaissance or experimental testing are not sufficient. On the other hand, numerical methods
105 use analytical procedures and apply computational frameworks such as the FE method. Using FE
106 is more feasible to generate more data points using tremendous ground motions and earthquake
107 scenarios. However, empirical or experimental data are still needed to define the damage states
108 needed to develop fragility curves numerically, and more importantly, to validate or calibrate the
109 used FE models. Several probability distribution functions have been considered for producing
110 fragility curves such as: normal functions, two combing normal functions (e.g. Anagnos 1999, Ang
111 et al. 1996), and lognormal functions (e.g. Hwang and Chou 1998, Shinozuka et al. 2007, Straub
112 and Der Kiureghian 2008). Typically, two parameters, the mean and deviation, are used to
113 determine the normal/lognormal fragility curve functions. The two combing normal functions
114 include one normal function for probabilities less than 0.5 and the other for probabilities greater
115 than 0.5. Four parameters are used for defining these combing normal functions, which include:
116 (1) the minimum PGA for the onset of damage, (2) PGA at the 16th damage percentile, (3) PGA at
117 the 50th damage percentile, and (4) PGA at the 84th damage percentile (Anagnos 1999). Moreover,
118 Bayesian methods, the Cornell method, and Boolean approaches have been also used. Most studies

4
119 chose the PGA for IM definition and fewer studies considered spectral acceleration (SA) instead.
120 For instance, Vanzi (1996) studied the fragility of a 420 kV circuit breaker using SA as IM. For
121 cases of electrical equipment, some researches set binary damage states, i.e. normal operation or
122 failure (e.g. Hwang and Chou 1998, Straub and Der Kiureghian 2008, Vanzi 1996). Others define
123 more detailed damage states or indices such as failure based on power imbalance, abnormal
124 voltage, unstable condition and operational power interruption (e.g., Anagnos 1999, Shinozuka et
125 al. 2007, Anagnos and Ostrom 2000), or classify it differently as none, slight, minor, moderate,
126 extensive and complete damage (e.g. Giovinazzi and King 2009).
127 Previous studies that varied in one of more of the above outlined features of fragility analysis
128 are presented here as examples. Ang (1996) assessed the electric power transmission systems using
129 synthetic and stochastic definition for the earthquake ground motions combined with seismic
130 capacities of electrical equipment as determined from available test data and simple modeling.
131 Wilcoski and Smith (1997) conducted fragility tests for power transformer bushings. Hwang and
132 Huo (1998) presented a fragility analysis case study for Substation 21 in Memphis using analytical
133 approach that did not consider actual earthquake records. Anagnos (1999) analyzed the fragility of
134 several substation components based on twelve California earthquakes. Anagnos and Ostrom
135 (2000) also presented examples of opinion-based fragility curves and proposed fragility curves for
136 500 kV circuit breakers and 230 kV horizontal-break disconnect switches. Rasulo et al. (2004)
137 evaluated seismic performance of a substation based on the 2002 Molise, Italy earthquake.
138 Shinozuka et al. (2007) investigated critical substation components, like transformers, circuit
139 breakers, switches and buses, using results acquired from 50 scenario earthquakes. Similarly,
140 Straub and Der Kiureghian (2008) presented fragility curves for type types of transformers and
141 circuit breakers. The studies that used empirical methods assembled the data based on the observed
142 numbers of failures in different substations during different earthquakes together with the
143 estimated PGA at each substation site. Recently, Zareei et al. (2016) presented the seismic failure
144 probability of a 400 kV power transformer using three-component original acceleration records. It
145 is worth mentioning that few studies extended the concepts of fragility analysis of electrical
146 equipment to hazards other than earthquakes. Shafieezadeh et al. (2014) studied the age-dependent
147 fragility models of electrical utility wood poles against extreme wind hazards, and Panteli and
148 Mancarella (2015) used the IEEE 6-bus reliability test system to illustrate its methodology with
149 three wind profiles with different densities: normal, high and extreme.

5
150 Most of the previous work summarized above is more of case studies for specific electrical
151 equipment or switches under specific scenarios and the analytical studies used simplified FE
152 models that lacked careful validation. No extensive or detailed work on ground motion variability,
153 operation configurations, equipment with different load capacity and structural systems has been
154 conducted. This study aims at filling some of the outlined knowledge gaps with focus on one type
155 of critical electrical substation equipment, namely disconnect switches.

156 MODEL DEVELOPMENT AND VERIFICATION

157 Two different voltage level disconnect switches, 230kV and 550kV, that are commonly used in
158 the Western United States are considered in this study. The two switches vary in size and support
159 structure configurations, and in turn, have different seismic response. The 230kV switches consists
160 of three phases, each phase is three post insulators connected using operation gate, and all phases
161 are typically supported on a three-dimensional braced steel frame (Fig. 1a). The 550kV switches
162 consist of only one three-insulator phase that is supported on a planer steel frame (Fig. 1b). Both
163 types of switches use porcelain post insulator that vary in size. Two operational configurations are
164 used in this study which consider open or closed gates. Fig. 1 shows both switches when
165 operational gates are open, i.e. indicating a configuration when the switch is disconnect from the
166 power line. To conduct all nonlinear analysis for the two switches, three-dimensional FE models
167 were developed using the commercial FE package ANSYS/LS DYNA (Release 18.0). The FE
168 models with the different switch components and for the two gate operational configurations are
169 shown in Figs. 2 and 3 for the 230kV and 550kV switches, respectively. Two nonlinear material
170 models were used. The bilinear kinematic hardening was used for the steel support structures and
171 insulator connections, and the brittle rupture model (*MAT_ADD_EROSION) was used for the
172 insulators porcelain core along with equivalent stress definition to determine when the insulator
173 fails. It is noted that porcelain insulators comprises steel caps where the porcelain parts are stacked
174 together. A metallic steel cap also contains a grout layer at the interface between porcelain core
175 and cap. Ideally, nonlinear material models are needed to capture the ceramic-steel-grout
176 interaction. However, the nonlinear behavior of these connections and joints were shown not to
177 affect the overall insulator macro model (Mosalam et al. 2012) and a linear model
178 (*MAT_ELASTIC) was used in this study for the metallic cap joints for simplicity. Viscous
179 damping equivalent to 2% damping ratio was used to define insulators inherent damping.

6
180
181 Figure 1. Two prototype disconnect switches: 230 kV open gate (left); 550 kV open gate (middle) and
182 closed gate (right) as setup for testing at UC Berkeley shake table (Mosalam et al. 2012)

183
184 Figure 2. FE model of a typical 230kV disconnect switch with the closed (left) and opened (right) gate
185 configurations and post insulators for different phases are identified

186
187 Figure 3. FE model of a typical 550kV disconnect switch with the closed (left) and opened (right) gate
188 configurations and the three post insulators for the switch phase are identified

7
189 A key aspect of this study is to develop accurate and high fidelity FE models that are validated
190 using various experimental tests to use for the fragility analysis. This study used results from about
191 12 component and two full switch dynamic tests (Fig. 1) that were previously conducted by, and
192 individual insulators fragility database that was assembled by one of the authors at the University
193 of California, Berkeley (Mosalam et al. 2012, Moustafa and Mosalam 2016a, 2016b). All material
194 and mechanical properties of the porcelain used in the post insulators, such as density, Young’s
195 modulus, Poisson’s ratio, etc., were acquired from porcelain cylinders material tests. To validate
196 first the individual insulators components of the FE models, lateral force-displacement relationship
197 until failure for both 230kV and 550kV insulators from pull tests and FE simulation were
198 compared. The comparison between the obtained force-displacement relationship from the tests
199 and FE are compared in Figs. 4 and 5 for the 230kV and 550kV insulators, respectively. The overall
200 nonlinear insulator behavior properly captured by the FE model provides confidence in using the
201 FE models to capture insulators failure for fragility assessment, and justifies that the simplification
202 of the porcelain-grout-steel interaction in the metallic cap regions as rigid joint is acceptable.

203
204 Figure 4. Force-displacement relationship until failure for 230kV insulator from FE and tests

205
206 Figure 5. Force-displacement relationship for 550kV support and insulator from FE and tests

8
207 The lateral stiffness simulated from the FE model for the 3D 230kV support structure is
208 11.23kN/mm in the out-of-plane and 36.48kN/mm in the in-plane directions, which compares to
209 11.40kN/mm and 35.21kN/mm, respectively, from experimental tests. For the 550 kV switch, the
210 lateral force-displacement relationship for the planer support structure (in-plane) is compared for
211 the FE and tests in Fig. 5. The natural frequencies were also compared for the individual
212 components and overall disconnect switch. The fundamental frequency for 230kV insulators
213 without any live parts varies from 12.55 Hz to 15.67 Hz which compares to 13.98 Hz from FE
214 model. When live parts are added to the 230KV insulator (jaw side as per Fig. 2), the natural
215 frequency captured from tests and the FE model is 5.13 Hz and 5.19 Hz, respectively. Moreover,
216 the natural frequency obtained for the 550kV rotating side insulator from both tests and FE is about
217 2.0 Hz. The good agreement between the FE and tests results for natural frequencies provides
218 confidence in using the model for capturing basic dynamic properties. However, to further validate
219 the models before conducting the fragility analysis, response histories and results from FE dynamic
220 analysis were compared against shake table tests conducted by Mosalam et al. (2012). For this part
221 of validation, an IEEE-spectrum compatible ground motion previously used for disconnect switch
222 shake table tests (Takhirov et al. 2005) was utilized. For the 550kV disconnect switch, the relative
223 displacement history between the rotating insulator top and support structure from a shake table
224 test of the full switch and FE analysis is compared for both closed and open gate configurations in
225 Fig. 6. Similarly, the 230kV jaw side insulator relative displacement response in an open gate
226 configuration is compared from a shake table test and FE analysis in Fig. 7. The comparison in
227 Figs. 6 and 7 further validate the FE models where similar response trends and peak response
228 values were obtained.

229
230 Figure 6. Relative displacement responses between rotating insulator top and shake table

9
231 
232 Figure 7. Lateral relative displacement responses of support from shake table test (Mosalam et al.
233 2012) and FE analysis (230kV DS Open)

234 FRAGILITY ASSESSMENT

235 DAMAGE STATES

236 To develop fragility curves, damage states should be accurately defined. For an individual
237 component in electrical equipment, multiple damage states (e.g. minor, moderate, or severe
238 damage) are not needed as the component whether fails or not during an earthquake. Accordingly,
239 a binary damage state is used in this study for the individual insulators, namely normal operation
240 or failure. A disconnect switch with several insulators, live parts, and a support structure could
241 have multiple damage states if failure of one or more components is observed. Moreover, the
242 overall switch might malfunction (e.g. gates fail to operate as required) without noticeable
243 structural damage, which might need to be reflected in the switch binary-damage-state-based
244 fragility functions. However, two main assumptions based on previous research and observations
245 from past earthquakes are considered here: (1) switches that pass seismic qualification tests are
246 likely to have functionality maintained up to the IEEE maximum considered earthquake level
247 before any structural failures (Mosalam et al. 2012); and (2) porcelain insulators are the most
248 fragile part of disconnect switches and one insulator failure results in the full switch
249 malfunctioning. Therefore, in this paper, the overall switch seismic response for fragility
250 development is expressed by one damage state (failure to operate) and this is identified by the first
251 insulator damage assuming functionality is maintained until a structural failure is observed.
252 However, the sequence of multiple component damage is still investigated here as discussed in the
253 next section), but with the objective of identifying which insulators are more critical. This is to
254 provide more insight into the maintenance activities, post-event inspection, or component
255 replacement such as replacing porcelain components with polymer ones as a retrofit measure for
256 instance.

10
257 To capture the damage and quantify the IM used for the fragility assessment, the incremental
258 dynamic analysis (IDA) method is used. Two acceleration-related IMs are used in this study as
259 previously mentioned, which are PGA and PRA at the insulators top. It is noted that one other
260 option for developing fragility curves is based on SA in the vicinity of equipment resonant
261 frequency (e.g. Vanzi 1996). For disconnect switches, SA-based fragilities are expected to show a
262 sharp change in the s-curve due to the brittle nature and binary damage state, which can be useful
263 for particular DS assessment. However, SA-based fragilities were not included in this study for
264 brevity and to provide more applicable fragility curves for regional assessments and augmented
265 grid nodes and substations fragilities. The IDA was carried out by applying a given ground motion
266 first scaled at 0.05g PGA then scale up with 0.05g increments until failure is captured and the
267 maximum PGA and the corresponding PRA that the disconnect switch can withstand are
268 determined. It is noted that all the ground motions used in this study are triaxial (more details
269 discussed in next subsection), so all three components of a ground motion were scaled similarly,
270 and the overall PGA at failure is taken as the largest value in all three directions. To identify when
271 damage happens in one or more insulators, the relative displacement between insulators top and
272 bottom was recorded and checked after every analysis case. For example, the out-of-plane relative
273 displacement response of the different 550kV switch insulators in an open gate configuration when
274 failure occurred is shown in Fig. 8.

275
276 Figure 8. Out-of-plane relative displacement response history between top and bottom of different
277 insulators and support structure of 550kV switch under one of the records scaled at PGA 0.85g

11
278 The sudden increase in the displacement values as shown in Fig. 8 indicates failure. The PGA
279 for this analysis case is 0.85g and the ground motion used is the IEEE-693-compatible record used
280 in the FE model validation and shake table comparisons from previous section. Thus, the 0.85g
281 PGA is considered the PGA level causing failure for the 550kV switch under that particular ground
282 motion, and in turn, 0.80g is considered the maximum PGA level the switch can withstand without
283 any damage. Note this example is only for illustration as this ground motion is not one of the 160
284 sets used in the fragility assessment. In general, not all the insulators fail at the same time or with
285 the same sequence as this varied from one analysis case to the other. However, examples of typical
286 failure modes as captured from the FE models are shown in Fig. 9 for the 550kV switch.

287
288 Figure 9. Examples of failure modes for 550kV FE model captured under different ground motions

289 To better illustrate that the sequence of insulators damage may vary according to the switch
290 type or gate configuration, Fig. 10 shows the relation between the PGA level and peak relative
291 displacement value for each individual insulator for four cases (two switch types × two gate
292 configurations). This relationship remains linear until failure happens. Fig. 10 summarizes analysis
293 cases using one ground motion only, but it illustrates that the sequence of insulators failure and
294 the PGA that corresponds to the first failure changes with the switch type or gate configuration.
295 For instance, the maximum PGA the 230kV switch withstands under the selected ground motion
296 case in a closed and open configurations is 0.90g and 0.70g, respectively. Also, when the PGA is
297 increased to 1.00g, 4 out of 9 insulators of the 230kV switch fail when the gate is closed but 8
298 insulators fail when the gate is opened. The ability of the FE models to capture the different
299 insulators failures and inform the fragility assessment in different operation configuration was not
300 featured in previous studies and is exclusively considered in this study.

12
301
302 Figure 10. Relationship between the out-of-plane peak relative displacement between the top and bottom
303 of different insulators and corresponding PGA values from the FE analysis
304 GROUND MOTIONS

305 Another feature of this study is to develop comprehensive fragility curves using different suites of
306 ground motions to better understand the effect of ground motion characteristics on seismic
307 performance of disconnect switches. Previous studies used different sets of ground motions for
308 lifelines and electrical equipment fragility analysis (e.g. Lee et al. 2005; Jayaram et al. 2011). In
309 this study, four comprehensive suites of ground motions that cover different scenarios and ground
310 motion characteristics as selected and readily provided by Baker et al. (2011) were used. These
311 ground motions are representative of Western United States and generated for locations throughout
312 California or other seismic active areas where seismic hazard is dominated by mid- to large-
313 magnitude crustal earthquakes at near to moderate distances. There are a total of 160 sets of triaxial
314 ground motions that have been divided into four suites or subgroups (40 sets per suite). The first
315 subgroup, designated as Set 1a, contains broad-band ground motions with relatively high
316 magnitude and close rupture distance (M=7, R=10km, soil site). Similarly, the second subgroup
317 (Set 1b) contains broad-band ground motions but at relatively far distance (M=6, R=25km, soil

13
318 site). Broad-band ground motions at rock site are included in the third subgroup (Set 2) for M=7,
319 R=10km, and rock site, and pulse-like ground motions (Set 3) are represented in the last subgroup.
320 The conducted IDA captured PGA and corresponding PRA at failure. Thus, unlike seismic
321 qualification testing, there was no need to scale the ground motions according to an IEEE
322 procedure for instance. However, all ground motions were scaled to 1.0g PGA for the convenience
323 of conducting the IDA starting at 0.5g PGA and with 0.05g increments. The normalized
324 acceleration spectra, i.e. scaled at 1.0 PGA herein, of all ground motions in the four different
325 subgroups are plotted in Fig. 11 using logarithmic scales.

326
327 Figure 11. Acceleration response spectra of the different suites of ground motions used for the disconnect
328 switches fragility assessment

329 FITTING FUNCTIONS

330 Seismic fragility functions or curves give the probability of exceedance of an identified response
331 of the structure at a specific performance level and for different values of a specific IM of the
332 earthquake. Such probability of exceedance of a specific objective response versus related IM
333 value typically has a lognormal distribution (Zareei et al. 2016, Baker 2015, Ghafory et al. 2011).
334 The lognormal distribution function is given in Equation (1) and it is the function used in this study
335 to develop the fragility curves.

14
ܺ
݈݊ ቀ ቁ
ሼ‫ܧ‬ȁ‫ ܯܫ‬ൌ ܺሽ ൌ ‫ ׎‬ቌ ߠ ቍ (1)
ߚ

336 where ሼ‫ܧ‬ȁ‫ ܯܫ‬ൌ ܺሽ is the probability of being at or exceeding ‫ ܯܫ‬ൌ ܺ. ‫׎‬ሺሻ is the standard
337 cumulative probability function, ߠ is the median of the fragility function, and ߚ is the standard
338 deviation. Calibrating Equation (1) for a given objective requires estimating ߠ and ߚ from
339 objective analysis results. Estimation of these parameters will be denoted as ߠ෠ and ߚመ . These
340 parameters are estimated through Equation (2) and Equation (3) so that the resulting fragility
341 function will be consistent with the observed data.

ͳ
Ž ߠ෠ ൌ ෍ Ž ‫ܯܫ‬௜ (2)
݊
௜ୀଵ


ͳ ‫ܯܫ‬௜ ଶ

ߚൌඩ ෍ሺŽሺ ሻሻ (3)
݊െͳ ߠ෠
௜ୀଵ

342 SYSTEM FRAGILITY CURVES

343 The basis for the fragility assessment previously discussed, including the damage states
344 identification, suites of ground motions, and fragility functions and IMs, was applied in this study
345 to develop fragility functions for 4 disconnect switch cases: two switch types (230kV and 550kV)
346 and two gate configurations (open and closed gates to disconnect or connect the switch to the
347 power line). The overall system fragilities based on all 160 ground motions combined is presented
348 in this section. More categorized fragility curves to understand the seismic response of individual
349 components of the switch and the effect of ground motion characteristics are presented and
350 discussed in the next two sections.
351 For all fragilities developed in this study, two IMs were used: the maximum PGA the switch
352 can withstand, and the PRA which the largest acceleration value typically observed at the top of
353 one of the insulators at failure. For a given ground motion, IDA is applied starting with the ground
354 motion scaled at 0.5g PGA then the scale is increased by 0.05g increments until failure is observed
355 and the corresponding PGA and PRA are recorded. This procedure was repeated for all 160 ground
356 motions and for the 4 different disconnect switch cases. For each case, 160 scatter points were
357 obtained and the fragility distribution function was fitted as illustrated in Fig. 12. The figure shows
358 the scattered analysis data points and corresponding fitted fragility curve when PGA is used as the

15
359 IM. It is noted that the figure is not meant to directly benefit the fragility analysis but only
360 illustrates how the fragility functions are fitted to individual or scatter data points obtained from
361 the IDA. This, in turn, demonstrates the validity of using lognormal distribution functions for
362 fitting fragility curves. The obtained system fragility curves based on PGA and PRA are plotted in
363 Fig. 13 for the four different analysis cases: (1) 230kV switch open gate; (2) 230kV switch closed
364 gate; (3) 550kV switch open gate; and (4) 550kV switch closed gate. The obtained system fragility
365 function parameters for the four cases are summarized in Table 1.

366
367 Figure 12. Scattered data and fitted fragility curves based on PGA

368
369 Figure 13. Fragility curves for different switch types and different configuration based on PGA and PRA

370 According to the data above, it can be seen that the switch phase configuration, close or open,
371 has significant influence on the 230kV disconnect switch seismic performance and fragility, but
372 only a small influence in the case of 550kV switch. At 50% probability of exceedance, the median

16
373 PGA for 230kV varies from 0.74g for open gate configuration to almost double the value for the
374 closed gate configuration. This means that the 230kV switch is more vulnerable when the operation
375 gate is opened, i.e. the switch is disconnected from the power line, which agrees with previous
376 research findings (e.g. Mosalam et al. 2012). On the other hand, the planer 550kV switch is found
377 to be less sensitive to the operation configuration. Only 6% difference in the median PGA value is
378 observed from the 550kV fragility curves with the open configuration still more vulnerable. This
379 is attributed to the nature of the 550kV switch support structure that is more flexible in the out-of-
380 plane direction regardless of the gate configuration. The figures and table above also indicate that
381 the PRA-based fragility curves are less sensitive to the gate configuration. The median PRA varied
382 only from 5.61g to 5.62g for the 230kV switch, and from 4.47g to 4.62g for the 550kV switch.
383 Table 1 – Parameters of system fragility curves based on PGA and PRA (all values are in g)
PGA used as IM PRA used as IM
Median (ș) Mean St.Dev (ȕ) Median (ș) Mean St.Dev (ȕ)
230kV Close 1.10 0.105 0.346 5.47 1.727 0.167
230kV Open 0.70 -0.298 0.372 5.74 1.724 0.216
550kV Close 1.10 0.101 0.403 4.39 1.531 0.260
550kV Open 1.05 0.051 0.385 4.32 1.498 0.281

384 SYSTEM VS. COMPONENT FRAGILITY

385 It is important to further understand the seismic performance of the individual components of the
386 two commonly used disconnect switches considered in this study to inform maintenance activities,
387 post-event inspections, retrofit measures and prioritization, or possibly enhance future designs.
388 The IDA was carried out for each analysis case until failure of each component, mainly individual
389 porcelain post insulators and the support structure, was captured. It is noted that the overall system
390 fragilities shown before were defined according to the first insulator failure, i.e. when the
391 disconnect switch is considered malfunctioning. However, additional fragility functions and
392 curves for components were calculated and presented here. One of the analysis assumptions
393 mentioned before is confirmed in this part of the study, which is demonstrating that the failure of
394 disconnect switches is governed by insulators damage rather than the support structure. For this
395 purpose, the fragility curves for the 550kV switch individual insulators and support structure were
396 determined. Nonlinear material models were used for both porcelain for the insulators and steel
397 for the support structure as previously discussed, and the limit state considered for the support
398 structure fragility was defined empirically based on engineering judgment as 2% residual drift
399 ratio. Component fragility curves (using PGA as IM) for the 550kV switch in open and closed gate

17
400 configuration are shown in Fig. 14. It can be seen from the figure that the median PGA value
401 corresponding to 50% probability of exceedance for the failure of the support structure is about 5
402 times higher than the insulators. Accordingly, the focus in the next part is the individual insulators
403 only. The goal is to have a better understanding of which insulators and in what gate configurations
404 are more vulnerable using the extensive probabilistic fragility assessment. This can help provide
405 more attention to the critical insulators as part of the design and maintenance activities.

406
407 Figure 14. Component fragility curves for the 550kV switch support structure and individual insulators in
408 the closed (left) and open (right) gate configurations

409 All the developed insulators component fragility curves for both disconnect switches are
410 plotted in Fig. 15. System fragility curves are also plotted in the same figures for comparison. A
411 summary of the fragility functions parameters for the component fragility curves is listed in Table
412 2. From the curves and the parameters table, it is observed that the probability of failure of a
413 specific component failure at a certain PGA level is generally larger than that of the overall system.
414 This is attributed to the fact that no specific insulator always fail first from the analysis, and in turn,
415 system fragilities dictated by first failure render a disconnect switch malfunctioning at higher risk
416 than individual component failure. The median PGA between the first and last insulator failure
417 varies more for the 230kV switch than the 550kV switch because of the planer support structure
418 that expose all 550kV insulators to similar seismic demands in the out-of-plane direction. The
419 vulnerability of individual insulators changes according to the operational configuration of the
420 switch. For instance, the Jaw side insulator of outermost phase of the 230kV is the most vulnerable
421 when all gates are closed while the rotating insulator is at the highest risk when gates are opened.
422 Also the median PGA at 50% probability of failure for the most and least vulnerable insulators
423 varies by about 50% and 55% when the gates are closed and opened, respectively. This is a change
424 from 0.86g to 1.31g for the open gate configuration (generally more vulnerable), for instance.

18
425
426 Figure 15. Individual insulators (component) fragility curves for all switches cases

427 Table 2 – Fragility functions parameters for insulators component (all values in g)
Insulators Closed gate configuration Open gate configuration
230kV switch Median (ș) Mean St.Dev (ȕ) Median (ș) Mean St.Dev (ȕ)
Jaw 1 1.15 0.146 0.370 1.05 0.083 0.397
Rot. 1 1.70 0.570 0.287 0.85 -0.147 0.452
Rig. 1 1.70 0.570 0.287 1.05 0.048 0.405
Jaw 2 1.35 0.300 0.356 1.30 0.270 0.358
Rot. 2 1.75 0.588 0.306 0.95 -0.007 0.444
Rig. 2 1.75 0.588 0.306 1.20 0.226 0.380
Jaw 3 1.30 0.268 0.339 1.10 0.088 0.390
Rot. 3 1.25 0.244 0.365 1.15 0.145 0.365
Rig. 3 1.25 0.244 0.365 1.15 0.155 0.375
550kV switch Median (g) Mean (g) StDev Median (g) Mean (g) StDev
Jaw 1 1.10 0.114 0.415 1.10 0.114 0.385
Rot. 1 1.25 0.226 0.404 1.20 0.259 0.448
Rig. 1 1.20 0.215 0.402 1.20 0.255 0.443
428
429 The variation in the median PGA is less pronounced in the 550kV switch. The range between
430 median PGA for most and least vulnerable insulators is from 1.12g to 1.25g when the gate is
431 closed, and 1.12g to 1.30g when the gate is opened. For the 230kV switch, the vulnerability of the
432 rotating insulator in each phase increases more than other insulators when the operation gates are
433 open. However, for the 550kV switch, the connected rotating and rigid insulators are always
434 coupled and showed almost exact fragilities that is less vulnerable than jaw-side insulators whether
435 the operation gate is opened or closed. It is worth noting that the variation in the seismic
436 performance of individual insulators is expected due to the change in the overall structural dynamic

19
437 characteristics and frequencies after every failing insulator. Finally, to demonstrate how the PGA
438 that corresponds to specific insulator failure or overall system first failure varied from one ground
439 motion to another, Table 3 shows a sample of this obtained data for the case of 230kV switch in a
440 close gate under the 40 individual ground motions used in Set 1a.
441 Table 3 – Sample comparison of the PGA at failure for individual components and the overall system
442 from the 40 different ground motions in Set 1a (all values are in g)
230kV Closed Individual components median PGA System
Set 1a GM# 1 2 3 4 5 6 7 8 9 PGA PRA
1 2.40 2.95 2.95 2.60 2.80 2.80 2.20 2.55 2.55 2.20 5.50
2 1.10 1.65 1.65 1.30 1.80 1.80 1.05 1.20 1.20 1.05 5.37
3 1.30 1.70 1.70 1.45 2.10 2.10 1.25 1.40 1.40 1.25 5.76
4 1.25 1.50 1.50 1.35 1.85 1.85 1.35 1.30 1.30 1.25 5.69
5 1.25 1.50 1.50 1.45 1.85 1.85 1.45 1.40 1.40 1.25 4.69
6 0.75 1.55 1.55 0.90 1.35 1.35 0.90 0.85 0.85 0.75 4.68
7 2.45 2.45 2.45 2.50 2.60 2.60 1.80 2.45 2.45 1.80 4.62
8 1.20 1.65 1.65 1.20 2.15 2.15 1.30 1.20 1.20 1.20 5.88
9 1.05 1.90 1.90 1.45 2.00 2.00 1.75 1.90 1.90 1.05 5.69
10 1.10 1.40 1.40 1.00 1.40 1.40 0.90 1.30 1.30 0.90 4.79
11 1.40 1.75 1.75 1.45 1.60 1.60 1.45 1.30 1.30 1.30 5.91
12 0.90 1.90 1.90 1.30 1.35 1.35 0.90 1.05 1.05 0.90 4.79
13 1.10 1.70 1.70 1.40 2.35 2.35 1.45 1.30 1.30 1.10 4.55
14 1.20 1.65 1.65 1.40 1.45 1.45 1.20 1.20 1.20 1.20 5.67
15 1.15 1.80 1.80 1.35 1.55 1.55 1.20 1.30 1.30 1.15 6.19
16 0.65 1.25 1.25 0.70 1.25 1.25 0.70 0.65 0.65 0.65 5.80
17 1.40 1.60 1.60 1.45 1.60 1.60 1.30 1.30 1.30 1.30 8.17
18 1.70 2.05 2.05 1.90 1.95 1.95 1.70 1.60 1.60 1.60 6.93
19 1.40 1.90 1.90 1.45 1.95 1.95 1.60 1.85 1.85 1.40 5.60
20 1.10 1.45 1.45 1.20 1.55 1.55 1.10 1.20 1.20 1.10 6.16
21 1.10 1.45 1.45 1.10 1.40 1.40 1.10 1.10 1.10 1.10 5.28
22 1.90 1.90 1.90 2.50 2.50 2.50 1.25 1.90 1.90 1.25 5.00
23 1.10 1.60 1.60 1.10 1.60 1.60 0.95 1.20 1.20 0.95 5.41
24 1.40 3.35 3.35 1.85 2.55 2.55 2.00 1.65 1.65 1.40 6.95
25 1.50 1.50 1.70 1.80 1.80 1.80 1.70 1.70 1.70 1.50 5.83
26 0.75 1.65 1.65 0.95 1.70 1.70 1.15 0.80 0.80 0.75 6.19
27 1.35 1.55 1.55 1.30 1.45 1.45 1.15 1.30 1.30 1.15 5.48
28 1.05 1.60 1.60 1.05 1.45 1.45 1.30 1.10 1.10 1.05 6.61
29 1.50 1.85 1.85 1.60 1.85 1.85 1.25 1.65 1.65 1.25 5.04
30 1.90 1.90 1.90 1.90 1.90 1.90 1.60 1.60 1.60 1.60 5.62
31 0.95 1.30 1.30 1.25 1.30 1.30 1.05 1.15 1.15 0.95 5.01
32 0.70 1.35 1.35 0.90 1.35 1.35 0.95 0.80 0.80 0.70 4.91
33 1.20 1.95 1.95 1.50 1.95 1.95 1.05 1.25 1.25 1.05 5.21
34 2.00 2.00 2.00 2.45 2.45 2.45 2.40 1.95 1.95 1.95 5.36
35 1.20 3.45 3.45 1.50 3.40 3.40 1.90 3.40 3.40 1.20 5.27
36 2.35 2.35 2.35 2.60 2.60 2.60 2.25 2.25 2.25 2.25 5.62
37 0.65 1.25 1.25 0.90 1.50 1.50 1.10 0.75 0.75 0.65 5.13
38 0.55 1.10 1.10 0.75 0.95 0.95 0.80 0.60 0.60 0.55 5.67
39 1.30 1.85 1.85 1.10 1.85 1.85 1.10 1.30 1.30 1.10 5.64
40 0.80 1.60 1.60 0.95 1.35 1.35 0.85 0.90 0.90 0.80 5.02

20
443 EFFECT OF GROUND MOTION CHARACTERISTICS

444 The last part of this study is focused on investigating the effects of different ground motion
445 characteristics on the vulnerability and overall system fragilities for the different disconnect
446 switches cases. A total of 160 sets of ground motions were used first to develop the system
447 fragilities (Section 4 above). In this section, four subsets of the 160 ground motion records (40 per
448 subset as explained in a previous section) were used to develop distinct system fragilities. The four
449 subsets represent variation in soil type, distance from source, and pulse-like records as selected
450 and grouped by Baker et al. (2011). The effects of these characteristics on the disconnect switches
451 are discussed. The developed fragility curves and fragility functions parameters based on using the
452 PGA and PRA as IMs under each subset of ground motions are summarized in Figs. 16 and 17 and
453 Tables 4 and 5, respectively.

454
455 Figure 16. System fragility curves (based on PGA IM) under different ground motions sets

456 It can be observed from the PGA-based fragilities that the 230kV switch is more vulnerable under
457 rock site conditions and is least vulnerable to the pulse-like ground motions regardless of the gate
458 operation configuration. However, for the 550kV switch, the proximity of the switch from the
459 earthquake source is the most pronounced ground motion characteristic in affecting the seismic
460 performance. In other words, 550kV switches on soil sites are more vulnerable when the rupture

21
461 distance is within 10 km (Set 1a) versus up to 25 km rupture distance (Set 1b). A general
462 observation is that the effects of ground motions on both switches are not sensitive to the operation
463 configuration of the switch. Both PGA and PRA based fragility curves show similar trends for the
464 effects of ground motion characteristics. However, the deviation in the PRA-based curves is on
465 average smaller than that of the corresponding PGA-based curves.

466
467 Figure 17. System fragility curves (based on PRA IM) under different ground motions sets

468 Table 4 – Parameters of PGA fragilities for different ground motion sets (all values in g)
Set 1a Median (ș) Mean St.Dev (ȕ) Set 1b Median (ș) Mean St.Dev (ȕ)
230kV-C 1.15 0.125 0.319 230kV-C 1.00 0.033 0.323
230kV-O 0.68 -0.341 0.320 230kV-O 0.73 -0.304 0.390
550kV-C 1.08 0.025 0.314 550kV-C 1.23 0.266 0.447
550kV-O 0.93 -0.074 0.317 550kV-O 1.20 0.168 0.476
Set 2 Median (ș) Mean St.Dev (ȕ) Set 3 Median (ș) Mean St.Dev (ȕ)
230kV-C 0.90 -0.048 0.295 230kV-C 1.30 0.310 0.352
230kV-O 0.65 -0.407 0.316 230kV-O 0.88 -0.140 0.412
550kV-C 1.00 0.005 0.366 550kV-C 1.00 0.109 0.433
550kV-O 1.00 -0.017 0.308 550kV-O 1.10 0.127 0.375
469

22
470 Table 5 – Parameters of PRA fragilities for different ground motion sets (all values in g)
Set 1a Median (ș) Mean St.Dev (ȕ) Set 1b Median (ș) Mean St.Dev (ȕ)
230kV-C 5.55 1.710 0.122 230kV-C 5.39 1.715 0.140
230kV-O 5.09 1.654 0.198 230kV-O 5.84 1.744 0.180
550kV-C 4.37 1.478 0.173 550kV-C 4.75 1.635 0.312
550kV-O 4.02 1.415 0.214 550kV-O 4.59 1.569 0.343
Set 2 Median (ș) Mean St.Dev (ȕ) Set 3 Median (ș) Mean St.Dev (ȕ)
230kV-C 5.24 1.677 0.100 230kV-C 5.68 1.807 0.244
230kV-O 6.21 1.767 0.209 230kV-O 5.59 1.730 0.259
550kV-C 4.22 1.509 0.249 550kV-C 4.27 1.503 0.265
550kV-O 4.35 1.509 0.264 550kV-O 4.24 1.499 0.278

471 SUMMARY AND CONCLUSIONS


472 Electrical substations have experienced severe seismic damage, and failure of vital components of
473 high-voltage substations such as disconnect switches caused blackouts during past earthquakes.
474 The literature lacked comprehensive fragility functions and curves that represent the seismic
475 vulnerability of such components as needed for power grids reliability and power loss assessments
476 as it relates to infrastructure maintenance, post-event inspection, or retrofit decision making
477 frameworks. The objective of this study was to investigate the seismic performance and develop
478 comprehensive component and system fragility curves for disconnect switches under large number
479 of ground motions representative of Western United States. Carefully calibrated and validated FE
480 models against experimental data were used to conduct this fragility study for two switch types
481 (230kV and 550kV) and consider for the first time, the operation configurations of the disconnect
482 switches (open and closed gate to disconnect or connect the switch to the power line). A
483 comprehensive suite of 160 triaxial ground motions that comprise 4 subsets with different ground
484 motion characteristics was used in this study. Nonlinear incremental dynamic analysis was
485 conducted for all switch cases to capture individual components and deduce overall system failure
486 under all ground motions. Based on this analytical study, the following conclusions are made:
487 x High fidelity nonlinear FE models that are calibrated using material tests and experimental
488 data can confidently model static, dynamic, and brittle failure behavior of porcelain insulators
489 used in critical electrical equipment as demonstrated for disconnect switches. Simplified
490 models could be used to represent the linear response of insulators until the sudden brittle
491 failure occurs for fragility assessment. However, detailed nonlinear models are needed to
492 establish accurate system and individual component fragilities, and identify the most critical
493 insulators in the different operation configurations.

23
494 x The seismic performance of disconnect switches can significantly vary when the operation
495 gates are open to disconnect the switch from the power line. This effect is more pronounced in
496 three-phase 230kV switches more than single-phase 550kV switches. The gates usually
497 connect the jaw side insulators to the already-connected rotating and rigid insulators.
498 Accordingly, rotating and rigid insulators show similar seismic fragilities but different from
499 the jaw side, especially when the operational gate is opened.
500 x Comprehensive PGA and PRA-based system fragility curves and functions for two common
501 types of disconnect switches in Western United States were developed for two operation
502 configurations when switches are connected or disconnected from the power line. Both types
503 of fragility curves have overall similar trends. However, PGA-based fragility curves highlight
504 more the effects of ground motions characteristics and operation gate configurations. Moreover,
505 the PGA-based system fragilities can be readily used by power utilities and augmented into
506 grid nodes and overall substation fragilities to conduct regional risk assessments and reliability
507 analysis for decision making frameworks, which inform maintenance activities, post-event
508 inspection, and retrofit measures of interconnected power grids.
509 x Individual component fragilities were developed and presented in this paper. This part of the
510 study confirmed that support structures are about 5 times less likely to fail than individual
511 insulators. The probability of failure of a specific switch component is generally less than the
512 overall switch failure at a given PGA intensity level. This is because no specific insulator
513 always breaks first, i.e. different insulators could fail at different times and a switch is
514 malfunctioning with the first major component failure.
515 x Seismic fragilities of disconnect switches were found to be affected by some of the
516 characteristics of ground motions such as site (rock vs. soil) and distance to rupture. However,
517 other characteristics like pulse-like ground motions do not seem to affect the switches seismic
518 fragilities. This can help inform future retrofit strategies and new designs, e.g. using polymer
519 insulators instead of porcelain, to mitigate the adverse effects of rock sites for instance.

520 ACKNOWLEDGEMENTS

521 This work is supported by Earthquake Research Fund of CEA (201508023), National Natural
522 Science Foundation of China (51478442), Program for Innovative Research Team in China
523 Earthquake Administration.

24
524 REFERENCES

525 Anagnos, T. (1999). Development of an electrical substation equipment performance database for
526 evaluation of equipment fragilities. Pacific Earthquake Engineering Research Center.
527 Anagnos, T., & Ostrom, D. K. (2000). Electrical substation equipment damage database for updating
528 fragility estimates. Proceedings of 12th World Conference on Earthquake Engineering, Auckland, New
529 Zeland, 30 January - 4 February, 2000.
530 Ang, A. S., Pires, J. A., & Villaverde, R. (1996). A model for the seismic reliability assessment of electric
531 power transmission systems. Reliability Engineering & System Safety, 51(1), 7-22.
532 Araneda, J. C., Rudnick, H., Mocarquer, S., & Miquel, P. (2010). Lessons from the 2010 Chilean earthquake
533 and its impact on electricity supply. In Power System Technology (POWERCON), 2010 International
534 Conference on (pp. 1-7). IEEE.
535 Baker, J. W., Lin, T., Shahi, S. K., & Jayaram, N. (2011). New ground motion selection procedures and
536 selected motions for the PEER transportation research program. Peer Report 2011, 3.
537 Baker, J. W. (2015). Efficient analytical fragility function fitting using dynamic structural
538 analysis. Earthquake Spectra, 31(1), 579-599.
539 Chang, L., & Wu, Z. (2011). Performance and reliability of electrical power grids under cascading
540 failures. International Journal of Electrical Power & Energy Systems, 33(8), 1410-1419.
541 Dueñas-Osorio, L., Craig, J. I., & Goodno, B. J. (2007a). Seismic response of critical interdependent
542 networks. Earthquake engineering & structural dynamics, 36(2), 285-306.
543 Dueñas-Osorio, L., Craig, J. I., Goodno, B. J., & Bostrom, A. (2007b). Interdependent response of
544 networked systems. Journal of Infrastructure Systems, 13(3), 185-194.
545 Dueñas-Osorio, L., & Kwasinski, A. (2012). Quantification of lifeline system interdependencies after the
546 27 February 2010 Mw 8.8 offshore Maule, Chile, earthquake. Earthquake Spectra, 28(S1), S581-S603.
547 Ghafory-Ashtiany, M., Mousavi, M., & Azarbakht, A. (2011). Strong ground motion record selection for
548 the reliable prediction of the mean seismic collapse capacity of a structure group. Earthquake
549 Engineering & Structural Dynamics, 40(6), 691-708.
550 Gilani, A. S., Whittaker, A. S., Fenves, G. L., & Fujisaki, E. (1999). Seismic evaluation of 550 kV porcelain
551 transformer bearings. Pacific Earthquake Engineering Research Center
552 Giovinazzi, S., & King, A. (2009). Estimating seismic impacts on lifelines: an international review for
553 RiskScape.
554 Hwang, H. H., & Chou, T. (1998). Evaluation of seismic performance of an electric substation using event
555 tree/fault tree technique. Probabilistic Engineering Mechanics, 13(2), 117-124.
556 Hwang, H. H., & Huo, J. R. (1998). Seismic fragility analysis of electric substation equipment and
557 structures. Probabilistic Engineering Mechanics, 13(2), 107-116.

25
558 IEEE Recommended Practice for Seismic Design of Substations, IEEE Standard 693-2005, Institute of
559 Electrical and Electronics Engineers, 2005.
560 Jayaram, N., Lin, T., & Baker, J. W. (2011). A computationally efficient ground-motion selection algorithm
561 for matching a target response spectrum mean and variance. Earthquake Spectra, 27(3), 797-815.
562 Kaynia, A. M., Taucer, F., & Hancilar, U. (Eds.). (2013). Guidelines for deriving seismic fragility functions
563 of elements at risk: Buildings, lifelines, transportation networks and critical facilities. Publications
564 Office.
565 Khalvati, A. H., & Hosseini, M. (2009). Seismic Performance of Electrical Substations' Equipments in Bam
566 Earthquake (Iran 2003). In TCLEE 2009: Lifeline Earthquake Engineering in a Multihazard
567 Environment (pp. 1-8).
568 Knight, B. T., & Kempner, L. (2009). Seismic Vulnerabilities and Retrofit of High-Voltage Electrical
569 Substation Facilities. Technical Council on Lifeline Earthquake Engineering Conference (pp.1-12).
570 Lee, J., Graf, W., Somerville, P., O’Rourke, T. D., & Shinozuka, M. (2005). Development of earthquake
571 scenarios for use in earthquake risk analysis for lifeline systems. Report for the Los Angeles
572 Department of Water and Power.
573 Massie, A., & Watson, N. R. (2011). Impact of the Christchurch earthquakes on the electrical power system
574 infrastructure. Bulletin of the New Zealand Society for Earthquake Engineering, 44(4), 425-430.
575 Mosalam, K. M., Moustafa, M. A., Günay, M. S., Triki, I., & Takhirov, S. (2012). Seismic performance of
576 substation insulator posts for vertical-break disconnect switches. California Energy Commission
577 Publication Number: CEC-500-2012.
578 Moustafa, M. A., & Mosalam, K. M. (2016a). Substructured dynamic testing of substation disconnect
579 switches. Earthquake Spectra, 32(1), 567-589.
580 Moustafa, M. A., & Mosalam, K. M. (2016b). Structural performance of porcelain and polymer post
581 insulators in high voltage electrical switches. Journal of Performance of Constructed Facilities, 30(5).
582 Nielson, B. G., & DesRoches, R. (2007). Analytical seismic fragility curves for typical bridges in the central
583 and southeastern United States. Earthquake Spectra, 23(3), 615-633.
584 Panteli, M., & Mancarella, P. (2015). Modeling and evaluating the resilience of critical electrical power
585 infrastructure to extreme weather events. IEEE Systems Journal.
586 Paolacci, F., & Giannini, R. (2005). Evaluation of seismic fragility of electrical insulators. In 9th
587 International Conference on Structural Safety and Reliability.
588 Paolacci, F., & Giannini, R. (2009). Seismic reliability assessment of a high-voltage disconnect switch
589 using an effective fragility analysis. Journal of Earthquake Engineering, 13(2), 217-235.
590 Paolacci, F., Giannini, R., Alessandri, S., & De Felice, G. (2014). Seismic vulnerability assessment of a
591 high voltage disconnect switch. Soil Dynamics and Earthquake Engineering, 67, 198-207.

26
592 Qiang, X., & Ruiyuan, Z. (2011). Damage to electric power grid infrastructure caused by natural disasters
593 in China. IEEE Power and Energy Magazine, 9(2), 28-36.
594 Rasulo, A., Goretti, A., & Nuti, C. (2004). Performance of lifelines during the 2002 Molise, Italy,
595 earthquake. Earthquake Spectra, 20(S1), S301-S314.
596 Reinhorn, A. M., Oikonomou, K., Roh, H., Schiff, A., & Kempner, Jr.(Leon). (2011). Modeling and seismic
597 performance evaluation of high voltage transformers and bushings. MCEER.
598 Rota, M., Penna, A., & Magenes, G. (2010). A methodology for deriving analytical fragility curves for
599 masonry buildings based on stochastic nonlinear analyses. Engineering Structures, 32(5), 1312-1323.
600 Schiff, Anshel J., and Alex K. Tang, eds. Chi-Chi, Taiwan, earthquake of September 21, 1999: lifeline
601 performance. Vol. 18. ASCE Publications, 2000.
602 Shafieezadeh, A., Onyewuchi, U. P., Begovic, M. M., & DesRoches, R. (2014). Age-dependent fragility
603 models of utility wood poles in power distribution networks against extreme wind hazards. IEEE
604 Transactions on Power Delivery, 29(1), 131-139.
605 Shinozuka, M., Dong, X., Chen, T. C., & Jin, X. (2007). Seismic performance of electric transmission
606 network under component failures. Earthquake engineering & structural dynamics, 36(2), 227-244.
607 Shinozuka, M., Feng, M., Dong, X., Chang, S. E., Cheng, T. C., Jin, X., & Saadeghvaziri, M. A. (2003).
608 Advances in seismic performance evaluation of power systems. Research Progress and
609 Accomplishments 2001-2003, 1-16.
610 Straub, D., & Der Kiureghian, A. (2008). Improved seismic fragility modeling from empirical
611 data. Structural safety, 30(4), 320-336.
612 Takhirov S. (2008). Seismic Qualification Report on EC-1, P and EV-1 Types of 245-kV Disconnect
613 Switches with EVG-1 Grounding Switch Installed. Report No. SS-SCE-2008, Earthquake Engineering
614 Research Center, University of California, Berkeley.
615 Takhirov, S., Fenves, G. L., Fujisaki, E., & Clyde, D. (2005). Ground motions for earthquake simulator
616 qualification of electrical substation equipment. Pacific Earthquake Engineering Research Center.
617 Takhirov, S., & Gilani, A. S. J. (2009). Earthquake Performance of High Voltage Electric Components and
618 New Standards for Seismic Qualification. Technical Council on Lifeline Earthquake Engineering
619 Conference (pp.1-11).
620 Thornberry L.R. and Hardy R. (1997). Seismic Simulation Test Program on Type CGCB-G 230 kV
621 Disconnect Switch. Report No. 45998-1. Wyle Laboratories Scientific Services and Systems Group:
622 Huntsville, Alabama.
623 Vanzi, I. (1996). Seismic reliability of electric power networks: methodology and application. Structural
624 Safety, 18(4), 311-327.
625 Vanzi, I., Bettinali, F., & Sigismondo, S. (2005). Fragility curves of electric substation equipment via the
626 Cornell method. In 9th International Conference on Structural Safety and Reliability, pp. 19-23.

27
627 Wilcoski, J., & Smith, S. J. (1997). Fragility Testing of a Power Transformer Bushing; Demonstration of
628 CERL Equipment Fragility and Protection Procedure (No. CERL/TR-97/57). Construction Engineering
629 Research Lab (Army) Champaign, IL.
630 Zareei, S. A., Hosseini, M., & Ghafory-Ashtiany, M. (2016). Seismic failure probability of a 400kV power
631 transformer using analytical fragility curves. Engineering Failure Analysis, 70, 273-289.

28

View publication stats

You might also like